ratio = ||descriptor1 - best_match|| / ||descriptor1 - second_best_match||
if ratio < 0.75:  # threshold típico
    accept_match()

# Librerías para procesamiento de imágenes y visión computacional
import cv2
import numpy as np

# Librerías para visualización
import matplotlib.pyplot as plt
import seaborn as sns

# Librerías para descarga de archivos
import gdown
import os
from pathlib import Path

# Configuración de visualización
plt.style.use('seaborn-v0_8-darkgrid')
sns.set_palette("husl")
%matplotlib inline

print("✅ Librerías importadas correctamente")
print(f"OpenCV version: {cv2.__version__}")

✅ Librerías importadas correctamente
OpenCV version: 4.12.0

# Diccionario con las imágenes y sus IDs de Google Drive
IMAGE_DATASET = {
    '01_Wall-E_Front_Environment1.jpg': '1qXaPcFsA9w4ce7fVvXkvauFiOKPrOiB8',
    '02_Wall-E_Right_Side_Environment1.jpg': '1O7v9FJOuynI1gW28FR3mJawKMxHjaIg2',
    '03_Wall-E_Right_Side_Environment2.jpg': '1aNybti8wTIswStoUcvDaf6ss6KhTljh7',
    '04_Wall-E_Left_Side_Environment2.jpg': '1ALYDEu_36moDBnAuOONAuXCflc8pUUKK',
    '05_Wall-E_Zoom_Out_Up_Environment2.jpg': '1Wzrj4FhvWGt9xI7kymqG08G2jmJCc_rO',
    '06_Wall-E_Close_Up_Front_Environment2.jpg': '1BDq3VTfukq1CeyS-GgwhgTTvR4-78TtV',
    '07_Wall-E_Laydown_Environment2.jpg': '1gwoIszT95veji_ITv5O_Mk9wAgUOFUw5',
    '08_Wall-E_Upside_Down_Environment2.jpg': '1sPM-ZGM0PZONefeXtYysR3WvxJLuO-O4',
    '09_Wall-E_And_His_Friends_CuyInFront_Environment1.jpg': '1VKJWf6KoA2SPv_T5qFPJzGry5RlruPqz',
    '10_Wall-E_And_His_Friends_Free_Front_Environment1.jpg': '1SMEbXT9bLlzONqgT7IKU7J096dFnA2D2',
    '11_Wall-E_And_His_Friends_Right_Side_Behind_Environment1.jpg': '1ApvBwr8ox-IJAnGnQ_vpSCBfeJAdZCTi',
    '12_Wall-E_Laydown_withObjects_Environment1.jpg': '1vOkPFIsVJDhWtkXZlzneLU6-sn4u1aHj',
    '13_Wall-E_Front_Outside_Environment3.jpg': '1FRlGdGAJn26Wg5w5Qfx-ZsBO2ztsASxZ',
    '14_Wall-E_Front_BehindAmongUs_Enviroment2.jpg': '1dd8tAPz8Jqe_L0Liyyko5HRhewypJWb_',
    '15_PopeyePuzzle_Front.jpg': '1tjEmHfm3UZmIkt7blOgGcsJ-AyKjmUfk',
    '16_PopeyePuzzle_MorePuzzles.jpg': '1Dq2s1kKf244yRqTPr7ascnTo8LQ_Vj_C',
    'elon_1.jpg': '1IQegnvUTqsOo0BFlwrRZ0LA9vgoA5NY1',
    'elon_2.png': '1MkG74Tzwl8BTUw4f3C7jZTnjB0uVFc3R'
}

# Crear directorio para las imágenes
DATA_DIR = Path('data')
DATA_DIR.mkdir(exist_ok=True)

def download_images(image_dict: dict, output_dir: Path) -> None:
    """
    Descarga imágenes desde Google Drive usando gdown.

    Args:
        image_dict: Diccionario con nombres de archivo como keys e IDs de Drive como values
        output_dir: Directorio donde se guardarán las imágenes
    """
    for filename, file_id in image_dict.items():
        output_path = output_dir / filename

        if output_path.exists():
            print(f"⏭️  {filename} ya existe, omitiendo descarga")
            continue

        url = f'https://drive.google.com/uc?id={file_id}'
        print(f"📥 Descargando {filename}...")
        gdown.download(url, str(output_path), quiet=False)

    print("\n✅ Todas las imágenes descargadas correctamente")

# Descargar imágenes
download_images(IMAGE_DATASET, DATA_DIR)

📥 Descargando 01_Wall-E_Front_Environment1.jpg...

Downloading...
From: https://drive.google.com/uc?id=1qXaPcFsA9w4ce7fVvXkvauFiOKPrOiB8
To: /content/data/01_Wall-E_Front_Environment1.jpg
100%|██████████| 331k/331k [00:00<00:00, 35.5MB/s]

📥 Descargando 02_Wall-E_Right_Side_Environment1.jpg...

Downloading...
From: https://drive.google.com/uc?id=1O7v9FJOuynI1gW28FR3mJawKMxHjaIg2
To: /content/data/02_Wall-E_Right_Side_Environment1.jpg
100%|██████████| 306k/306k [00:00<00:00, 8.43MB/s]

📥 Descargando 03_Wall-E_Right_Side_Environment2.jpg...

Downloading...
From: https://drive.google.com/uc?id=1aNybti8wTIswStoUcvDaf6ss6KhTljh7
To: /content/data/03_Wall-E_Right_Side_Environment2.jpg
100%|██████████| 500k/500k [00:00<00:00, 14.1MB/s]

📥 Descargando 04_Wall-E_Left_Side_Environment2.jpg...

Downloading...
From: https://drive.google.com/uc?id=1ALYDEu_36moDBnAuOONAuXCflc8pUUKK
To: /content/data/04_Wall-E_Left_Side_Environment2.jpg
100%|██████████| 500k/500k [00:00<00:00, 9.01MB/s]

📥 Descargando 05_Wall-E_Zoom_Out_Up_Environment2.jpg...

Downloading...
From: https://drive.google.com/uc?id=1Wzrj4FhvWGt9xI7kymqG08G2jmJCc_rO
To: /content/data/05_Wall-E_Zoom_Out_Up_Environment2.jpg
100%|██████████| 696k/696k [00:00<00:00, 30.7MB/s]

def load_and_preprocess_image(image_path: str, max_width: int = 800,
                             convert_to_gray: bool = True) -> tuple:
    """
    Carga y preprocesa una imagen: redimensiona y convierte a escala de grises.

    Args:
        image_path: Ruta de la imagen
        max_width: Ancho máximo para redimensionar
        convert_to_gray: Si True, convierte a escala de grises

    Returns:
        Tupla (imagen_color, imagen_gris) o (imagen_color, None) si convert_to_gray=False
    """
    # Leer imagen
    img = cv2.imread(image_path)
    if img is None:
        raise FileNotFoundError(f"No se pudo cargar la imagen: {image_path}")

    # Redimensionar si es necesario
    height, width = img.shape[:2]
    if width > max_width:
        scale = max_width / width
        new_width = max_width
        new_height = int(height * scale)
        img = cv2.resize(img, (new_width, new_height), interpolation=cv2.INTER_AREA)

    # Convertir BGR a RGB para visualización correcta
    img_rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)

    # Convertir a escala de grises si se requiere
    img_gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY) if convert_to_gray else None

    return img_rgb, img_gray

def display_image_pair(img1: np.ndarray, img2: np.ndarray,
                       title1: str = "Imagen 1", title2: str = "Imagen 2",
                       figsize: tuple = (16, 6), cmap: str = None) -> None:
    """
    Visualiza dos imágenes lado a lado.

    Args:
        img1, img2: Imágenes a visualizar
        title1, title2: Títulos de las imágenes
        figsize: Tamaño de la figura
        cmap: Mapa de colores (ej. 'gray' para escala de grises)
    """
    fig, axes = plt.subplots(1, 2, figsize=figsize)

    axes[0].imshow(img1, cmap=cmap)
    axes[0].set_title(title1, fontsize=12, fontweight='bold')
    axes[0].axis('off')

    axes[1].imshow(img2, cmap=cmap)
    axes[1].set_title(title2, fontsize=12, fontweight='bold')
    axes[1].axis('off')

    plt.tight_layout()
    plt.show()

print("✅ Funciones de utilidad definidas")

✅ Funciones de utilidad definidas

# Cargar imagen de demostración
img_color, img_gray = load_and_preprocess_image('data/elon_1.jpg')

# Visualizar imagen original
display_image_pair(img_color, img_gray,
                   title1="Imagen Original (Color)",
                   title2="Imagen en Escala de Grises",
                   cmap='gray')

def create_transformed_image(img: np.ndarray, scale_levels: int = 2,
                            rotation_angle: float = 30) -> tuple:
    """
    Crea una versión transformada de la imagen con cambios de escala y rotación.

    Args:
        img: Imagen original
        scale_levels: Número de niveles de pirámide (escala)
        rotation_angle: Ángulo de rotación en grados

    Returns:
        Tupla (imagen_transformada_color, imagen_transformada_gray)
    """
    # Aplicar reducción de escala mediante pirámide gaussiana
    transformed = img.copy()
    for _ in range(scale_levels):
        transformed = cv2.pyrDown(transformed)

    # Aplicar rotación
    num_rows, num_cols = transformed.shape[:2]
    center = (num_cols / 2, num_rows / 2)
    rotation_matrix = cv2.getRotationMatrix2D(center, rotation_angle, 1.0)
    transformed = cv2.warpAffine(transformed, rotation_matrix, (num_cols, num_rows))

    # Convertir a escala de grises
    transformed_gray = cv2.cvtColor(transformed, cv2.COLOR_RGB2GRAY)

    return transformed, transformed_gray

# Crear imagen de prueba con transformaciones
test_img_color, test_img_gray = create_transformed_image(img_color,
                                                          scale_levels=2,
                                                          rotation_angle=30)

print(f"✅ Imagen transformada creada")
print(f"   - Escala reducida: 2 niveles de pirámide")
print(f"   - Rotación aplicada: 30 grados")
print(f"   - Tamaño original: {img_color.shape[:2]}")
print(f"   - Tamaño transformado: {test_img_color.shape[:2]}")

✅ Imagen transformada creada
   - Escala reducida: 2 niveles de pirámide
   - Rotación aplicada: 30 grados
   - Tamaño original: (357, 632)
   - Tamaño transformado: (90, 158)

display_image_pair(img_color, test_img_color,
                   title1="Imagen de Entrenamiento (Original)",
                   title2="Imagen de Prueba (Escalada + Rotada)",
                   figsize=(18, 7))

# Crear detector ORB
orb = cv2.ORB_create(nfeatures=1000)

# Detectar keypoints y computar descriptores
train_keypoints, train_descriptors = orb.detectAndCompute(img_color, None)
test_keypoints, test_descriptors = orb.detectAndCompute(test_img_gray, None)

print(f"🔍 Keypoints detectados:")
print(f"   - Imagen de entrenamiento: {len(train_keypoints)}")
print(f"   - Imagen de prueba: {len(test_keypoints)}")

# Visualizar keypoints
img_keypoints_simple = cv2.drawKeypoints(img_color, train_keypoints, None,
                                          color=(0, 255, 0))
img_keypoints_rich = cv2.drawKeypoints(img_color, train_keypoints, None,
                                        flags=cv2.DRAW_MATCHES_FLAGS_DRAW_RICH_KEYPOINTS)

display_image_pair(img_keypoints_rich, img_keypoints_simple,
                   title1="Keypoints con Tamaño y Orientación",
                   title2="Keypoints Simples",
                   figsize=(18, 7))

🔍 Keypoints detectados:
   - Imagen de entrenamiento: 960
   - Imagen de prueba: 39

# Crear Brute Force Matcher para descriptores binarios
bf_matcher = cv2.BFMatcher(cv2.NORM_HAMMING, crossCheck=True)

# Realizar matching
matches = bf_matcher.match(train_descriptors, test_descriptors)
matches = sorted(matches, key=lambda x: x.distance)

# Visualizar mejores matches
num_matches_to_show = min(50, len(matches))
img_matches = cv2.drawMatches(img_color, train_keypoints,
                              test_img_gray, test_keypoints,
                              matches[:num_matches_to_show], None,
                              flags=cv2.DrawMatchesFlags_NOT_DRAW_SINGLE_POINTS)

plt.figure(figsize=(18, 8))
plt.title(f'ORB - Mejores {num_matches_to_show} Matches', fontsize=14, fontweight='bold')
plt.imshow(img_matches)
plt.axis('off')
plt.tight_layout()
plt.show()

print(f"\n📊 Estadísticas de Matching ORB:")
print(f"   - Total de matches encontrados: {len(matches)}")
print(f"   - Distancia promedio (top 50): {np.mean([m.distance for m in matches[:50]]):.2f}")

📊 Estadísticas de Matching ORB:
   - Total de matches encontrados: 30
   - Distancia promedio (top 50): 54.07

# Cargar imagen y convertir a escala de grises
img_sift, gray_sift = load_and_preprocess_image('data/elon_1.jpg')

# Crear detector SIFT
sift = cv2.SIFT_create()

# Detectar keypoints y descriptores
keypoints_sift, descriptors_sift = sift.detectAndCompute(img_sift, None)

# Dibujar keypoints
img_with_keypoints = cv2.drawKeypoints(gray_sift, keypoints_sift, img_sift,
                                        flags=cv2.DRAW_MATCHES_FLAGS_DRAW_RICH_KEYPOINTS)

plt.figure(figsize=(14, 8))
plt.title(f'SIFT Keypoints Detectados: {len(keypoints_sift)}',
          fontsize=14, fontweight='bold')
plt.imshow(img_with_keypoints)
plt.axis('off')
plt.tight_layout()
plt.show()

print(f"🔍 Keypoints SIFT detectados: {len(keypoints_sift)}")

🔍 Keypoints SIFT detectados: 293

# Cargar par de imágenes
img1_color, img1_gray = load_and_preprocess_image('data/elon_1.jpg')
img2_color, img2_gray = load_and_preprocess_image('data/elon_2.png')

# Visualizar ambas imágenes
display_image_pair(img1_gray, img2_gray,
                   title1="Imagen 1 - Elon Musk",
                   title2="Imagen 2 - Elon Musk (Diferente Pose)",
                   cmap='gray')

# Detectar keypoints y descriptores en ambas imágenes
sift = cv2.SIFT_create()

kp1, desc1 = sift.detectAndCompute(img1_gray, None)
kp2, desc2 = sift.detectAndCompute(img2_gray, None)

print(f"🔍 Keypoints extraídos:")
print(f"   - Imagen 1: {len(kp1)} keypoints")
print(f"   - Imagen 2: {len(kp2)} keypoints")

🔍 Keypoints extraídos:
   - Imagen 1: 296 keypoints
   - Imagen 2: 390 keypoints

# Crear matcher con norma L2
bf = cv2.BFMatcher(cv2.NORM_L2, crossCheck=True)

# Realizar matching
matches = bf.match(desc1, desc2)
matches = sorted(matches, key=lambda x: x.distance)

# Visualizar top 50 matches
img_matches = cv2.drawMatches(img1_gray, kp1, img2_gray, kp2,
                              matches[:50], None,
                              flags=cv2.DrawMatchesFlags_NOT_DRAW_SINGLE_POINTS)

plt.figure(figsize=(18, 8))
plt.title('SIFT - Top 50 Matches', fontsize=14, fontweight='bold')
plt.imshow(img_matches, cmap='gray')
plt.axis('off')
plt.tight_layout()
plt.show()

print(f"\n📊 Total de matches encontrados: {len(matches)}")

📊 Total de matches encontrados: 102

class FeatureMatchingComparator:
    """
    Clase para comparar algoritmos de feature matching.
    Implementa Harris, SIFT y ORB con métricas de evaluación.
    """

    def __init__(self):
        self.metrics = {
            'image_pairs': [],
            'sift_50': [], 'sift_all': [],
            'orb_50': [], 'orb_all': [],
            'harris_50': [], 'harris_all': [],
            'keypoints_sift': [], 'keypoints_orb': [], 'keypoints_harris': [],
            'matches_sift': [], 'matches_orb': [], 'matches_harris': []
        }
        self.orb = cv2.ORB_create(nfeatures=1000)
        self.sift = cv2.SIFT_create()

    def _extract_harris_keypoints(self, img_gray: np.ndarray) -> list:
        """
        Extrae keypoints usando Harris Corner Detector.

        Args:
            img_gray: Imagen en escala de grises

        Returns:
            Lista de cv2.KeyPoint objetos
        """
        harris_corners = cv2.cornerHarris(img_gray, blockSize=2, ksize=3, k=0.04)
        harris_corners = cv2.dilate(harris_corners, None)

        # Umbral para seleccionar esquinas prominentes
        threshold = 0.01 * harris_corners.max()
        keypoint_coords = np.argwhere(harris_corners > threshold)

        # Convertir coordenadas a KeyPoint objects
        keypoints = [cv2.KeyPoint(float(pt[1]), float(pt[0]), 1)
                    for pt in keypoint_coords]

        return keypoints

    def _compute_average_distance(self, matches: list, num_matches: int) -> float:
        """
        Calcula distancia promedio de los primeros N matches.

        Args:
            matches: Lista de DMatch objects
            num_matches: Número de matches a considerar

        Returns:
            Distancia promedio
        """
        if not matches or num_matches <= 0:
            return 0.0

        subset = matches[:min(num_matches, len(matches))]
        return sum(m.distance for m in subset) / len(subset)

    def _format_image_pair_name(self, path1: str, path2: str) -> str:
        """
        Formatea nombres de archivos para etiquetas de gráficos.
        """
        name1 = Path(path1).stem[:15]
        name2 = Path(path2).stem[:15]
        return f"{name1}\nvs\n{name2}"

    def _plot_matches(self, title: str, img1: np.ndarray, kp1: list,
                     img2: np.ndarray, kp2: list, matches: list,
                     num_matches: int) -> None:
        """
        Visualiza matches entre dos imágenes.
        """
        matches_to_draw = matches[:min(num_matches, len(matches))]
        img_matches = cv2.drawMatches(img1, kp1, img2, kp2, matches_to_draw, None,
                                     flags=cv2.DrawMatchesFlags_NOT_DRAW_SINGLE_POINTS)

        plt.figure(figsize=(18, 6))
        plt.title(f'{title} - {len(matches_to_draw)} Matches',
                 fontsize=13, fontweight='bold')
        plt.imshow(img_matches, cmap='gray')
        plt.axis('off')
        plt.tight_layout()
        plt.show()

    def compare_algorithms(self, img1_path: str, img2_path: str,
                          visualize: bool = True) -> dict:
        """
        Compara Harris, SIFT y ORB en un par de imágenes.

        Args:
            img1_path: Ruta de primera imagen
            img2_path: Ruta de segunda imagen
            visualize: Si True, muestra visualizaciones

        Returns:
            Diccionario con métricas del par de imágenes
        """
        # Cargar imágenes
        _, img1_gray = load_and_preprocess_image(img1_path)
        _, img2_gray = load_and_preprocess_image(img2_path)

        pair_name = self._format_image_pair_name(img1_path, img2_path)

        if visualize:
            display_image_pair(img1_gray, img2_gray,
                             title1=Path(img1_path).name,
                             title2=Path(img2_path).name,
                             cmap='gray', figsize=(16, 5))

        results = {}

        # ========== HARRIS + ORB DESCRIPTORS ==========
        kp1_harris = self._extract_harris_keypoints(img1_gray)
        kp2_harris = self._extract_harris_keypoints(img2_gray)
        _, desc1_harris = self.orb.compute(img1_gray, kp1_harris)
        _, desc2_harris = self.orb.compute(img2_gray, kp2_harris)

        if desc1_harris is not None and desc2_harris is not None:
            bf_harris = cv2.BFMatcher(cv2.NORM_HAMMING, crossCheck=True)
            matches_harris = sorted(bf_harris.match(desc1_harris, desc2_harris),
                                   key=lambda x: x.distance)

            results['harris'] = {
                'keypoints': (len(kp1_harris), len(kp2_harris)),
                'matches': len(matches_harris),
                'avg_dist_50': self._compute_average_distance(matches_harris, 50),
                'avg_dist_all': self._compute_average_distance(matches_harris, len(matches_harris))
            }

            if visualize:
                self._plot_matches('Harris + ORB Descriptors', img1_gray, kp1_harris,
                                 img2_gray, kp2_harris, matches_harris, 50)

        # ========== SIFT ==========
        kp1_sift, desc1_sift = self.sift.detectAndCompute(img1_gray, None)
        kp2_sift, desc2_sift = self.sift.detectAndCompute(img2_gray, None)

        bf_sift = cv2.BFMatcher(cv2.NORM_L2, crossCheck=True)
        matches_sift = sorted(bf_sift.match(desc1_sift, desc2_sift),
                             key=lambda x: x.distance)

        results['sift'] = {
            'keypoints': (len(kp1_sift), len(kp2_sift)),
            'matches': len(matches_sift),
            'avg_dist_50': self._compute_average_distance(matches_sift, 50),
            'avg_dist_all': self._compute_average_distance(matches_sift, len(matches_sift))
        }

        if visualize:
            self._plot_matches('SIFT', img1_gray, kp1_sift,
                             img2_gray, kp2_sift, matches_sift, 50)

        # ========== ORB ==========
        kp1_orb, desc1_orb = self.orb.detectAndCompute(img1_gray, None)
        kp2_orb, desc2_orb = self.orb.detectAndCompute(img2_gray, None)

        bf_orb = cv2.BFMatcher(cv2.NORM_HAMMING, crossCheck=True)
        matches_orb = sorted(bf_orb.match(desc1_orb, desc2_orb),
                            key=lambda x: x.distance)

        results['orb'] = {
            'keypoints': (len(kp1_orb), len(kp2_orb)),
            'matches': len(matches_orb),
            'avg_dist_50': self._compute_average_distance(matches_orb, 50),
            'avg_dist_all': self._compute_average_distance(matches_orb, len(matches_orb))
        }

        if visualize:
            self._plot_matches('ORB', img1_gray, kp1_orb,
                             img2_gray, kp2_orb, matches_orb, 50)

        # Almacenar métricas
        self.metrics['image_pairs'].append(pair_name)
        self.metrics['sift_50'].append(results['sift']['avg_dist_50'])
        self.metrics['sift_all'].append(results['sift']['avg_dist_all'])
        self.metrics['orb_50'].append(results['orb']['avg_dist_50'])
        self.metrics['orb_all'].append(results['orb']['avg_dist_all'])
        self.metrics['harris_50'].append(results['harris']['avg_dist_50'])
        self.metrics['harris_all'].append(results['harris']['avg_dist_all'])

        self.metrics['keypoints_sift'].append(results['sift']['keypoints'][0])
        self.metrics['keypoints_orb'].append(results['orb']['keypoints'][0])
        self.metrics['keypoints_harris'].append(results['harris']['keypoints'][0])

        self.metrics['matches_sift'].append(results['sift']['matches'])
        self.metrics['matches_orb'].append(results['orb']['matches'])
        self.metrics['matches_harris'].append(results['harris']['matches'])

        # Imprimir resultados
        print(f"\n{'='*60}")
        print(f"📊 Resultados para: {Path(img1_path).name} vs {Path(img2_path).name}")
        print(f"{'='*60}")

        for algo_name, algo_results in results.items():
            print(f"\n🔹 {algo_name.upper()}:")
            print(f"   Keypoints detectados: Img1={algo_results['keypoints'][0]}, "
                  f"Img2={algo_results['keypoints'][1]}")
            print(f"   Total matches: {algo_results['matches']}")
            print(f"   Distancia promedio (50 mejores): {algo_results['avg_dist_50']:.2f}")
            print(f"   Distancia promedio (todos): {algo_results['avg_dist_all']:.2f}")

        return results

# Crear instancia del comparador
comparator = FeatureMatchingComparator()

print("✅ Clase FeatureMatchingComparator inicializada")

✅ Clase FeatureMatchingComparator inicializada

# Comparar imágenes de Elon Musk
results_elon = comparator.compare_algorithms('data/elon_1.jpg', 'data/elon_2.png')

============================================================
📊 Resultados para: elon_1.jpg vs elon_2.png
============================================================

🔹 HARRIS:
   Keypoints detectados: Img1=1125, Img2=1244
   Total matches: 152
   Distancia promedio (50 mejores): 17.26
   Distancia promedio (todos): 28.60

🔹 SIFT:
   Keypoints detectados: Img1=296, Img2=390
   Total matches: 102
   Distancia promedio (50 mejores): 204.85
   Distancia promedio (todos): 263.11

🔹 ORB:
   Keypoints detectados: Img1=967, Img2=999
   Total matches: 311
   Distancia promedio (50 mejores): 33.10
   Distancia promedio (todos): 49.56

# Wall-E: Mismo entorno, diferentes ángulos
comparator.compare_algorithms(
    'data/01_Wall-E_Front_Environment1.jpg',
    'data/02_Wall-E_Right_Side_Environment1.jpg'
)

============================================================
📊 Resultados para: 01_Wall-E_Front_Environment1.jpg vs 02_Wall-E_Right_Side_Environment1.jpg
============================================================

🔹 HARRIS:
   Keypoints detectados: Img1=7251, Img2=6978
   Total matches: 618
   Distancia promedio (50 mejores): 22.80
   Distancia promedio (todos): 48.06

🔹 SIFT:
   Keypoints detectados: Img1=1389, Img2=1290
   Total matches: 390
   Distancia promedio (50 mejores): 133.08
   Distancia promedio (todos): 275.55

🔹 ORB:
   Keypoints detectados: Img1=1000, Img2=1000
   Total matches: 256
   Distancia promedio (50 mejores): 39.00
   Distancia promedio (todos): 52.80

{'harris': {'keypoints': (7251, 6978),
  'matches': 618,
  'avg_dist_50': 22.8,
  'avg_dist_all': 48.06310679611651},
 'sift': {'keypoints': (1389, 1290),
  'matches': 390,
  'avg_dist_50': 133.08342834472657,
  'avg_dist_all': 275.55492819761616},
 'orb': {'keypoints': (1000, 1000),
  'matches': 256,
  'avg_dist_50': 39.0,
  'avg_dist_all': 52.80078125}}

# Wall-E: Rotación extrema (upside down)
comparator.compare_algorithms(
    'data/01_Wall-E_Front_Environment1.jpg',
    'data/08_Wall-E_Upside_Down_Environment2.jpg'
)

============================================================
📊 Resultados para: 01_Wall-E_Front_Environment1.jpg vs 08_Wall-E_Upside_Down_Environment2.jpg
============================================================

🔹 HARRIS:
   Keypoints detectados: Img1=7251, Img2=25636
   Total matches: 1135
   Distancia promedio (50 mejores): 24.44
   Distancia promedio (todos): 47.06

🔹 SIFT:
   Keypoints detectados: Img1=1389, Img2=4724
   Total matches: 622
   Distancia promedio (50 mejores): 108.51
   Distancia promedio (todos): 258.08

🔹 ORB:
   Keypoints detectados: Img1=1000, Img2=1000
   Total matches: 246
   Distancia promedio (50 mejores): 42.46
   Distancia promedio (todos): 56.77

{'harris': {'keypoints': (7251, 25636),
  'matches': 1135,
  'avg_dist_50': 24.44,
  'avg_dist_all': 47.05903083700441},
 'sift': {'keypoints': (1389, 4724),
  'matches': 622,
  'avg_dist_50': 108.50702880859374,
  'avg_dist_all': 258.07929614885825},
 'orb': {'keypoints': (1000, 1000),
  'matches': 246,
  'avg_dist_50': 42.46,
  'avg_dist_all': 56.77235772357724}}

# Wall-E: Cambio de escala (zoom out)
comparator.compare_algorithms(
    'data/01_Wall-E_Front_Environment1.jpg',
    'data/05_Wall-E_Zoom_Out_Up_Environment2.jpg'
)

============================================================
📊 Resultados para: 01_Wall-E_Front_Environment1.jpg vs 05_Wall-E_Zoom_Out_Up_Environment2.jpg
============================================================

🔹 HARRIS:
   Keypoints detectados: Img1=7251, Img2=99433
   Total matches: 1647
   Distancia promedio (50 mejores): 25.32
   Distancia promedio (todos): 49.22

🔹 SIFT:
   Keypoints detectados: Img1=1389, Img2=10632
   Total matches: 742
   Distancia promedio (50 mejores): 169.28
   Distancia promedio (todos): 283.92

🔹 ORB:
   Keypoints detectados: Img1=1000, Img2=1000
   Total matches: 239
   Distancia promedio (50 mejores): 52.40
   Distancia promedio (todos): 64.68

{'harris': {'keypoints': (7251, 99433),
  'matches': 1647,
  'avg_dist_50': 25.32,
  'avg_dist_all': 49.22465088038859},
 'sift': {'keypoints': (1389, 10632),
  'matches': 742,
  'avg_dist_50': 169.28126541137695,
  'avg_dist_all': 283.91926306341537},
 'orb': {'keypoints': (1000, 1000),
  'matches': 239,
  'avg_dist_50': 52.4,
  'avg_dist_all': 64.68200836820084}}

# Wall-E: Oclusión parcial con objetos
comparator.compare_algorithms(
    'data/01_Wall-E_Front_Environment1.jpg',
    'data/12_Wall-E_Laydown_withObjects_Environment1.jpg'
)

============================================================
📊 Resultados para: 01_Wall-E_Front_Environment1.jpg vs 12_Wall-E_Laydown_withObjects_Environment1.jpg
============================================================

🔹 HARRIS:
   Keypoints detectados: Img1=7251, Img2=29901
   Total matches: 1124
   Distancia promedio (50 mejores): 22.68
   Distancia promedio (todos): 45.84

🔹 SIFT:
   Keypoints detectados: Img1=1389, Img2=3384
   Total matches: 526
   Distancia promedio (50 mejores): 95.13
   Distancia promedio (todos): 253.26

🔹 ORB:
   Keypoints detectados: Img1=1000, Img2=1000
   Total matches: 233
   Distancia promedio (50 mejores): 43.80
   Distancia promedio (todos): 57.51

{'harris': {'keypoints': (7251, 29901),
  'matches': 1124,
  'avg_dist_50': 22.68,
  'avg_dist_all': 45.844306049822066},
 'sift': {'keypoints': (1389, 3384),
  'matches': 526,
  'avg_dist_50': 95.12756065368653,
  'avg_dist_all': 253.2586680002539},
 'orb': {'keypoints': (1000, 1000),
  'matches': 233,
  'avg_dist_50': 43.8,
  'avg_dist_all': 57.506437768240346}}

# Wall-E: Con otros objetos en la escena
comparator.compare_algorithms(
    'data/01_Wall-E_Front_Environment1.jpg',
    'data/09_Wall-E_And_His_Friends_CuyInFront_Environment1.jpg'
)

============================================================
📊 Resultados para: 01_Wall-E_Front_Environment1.jpg vs 09_Wall-E_And_His_Friends_CuyInFront_Environment1.jpg
============================================================

🔹 HARRIS:
   Keypoints detectados: Img1=7251, Img2=7799
   Total matches: 645
   Distancia promedio (50 mejores): 23.98
   Distancia promedio (todos): 43.46

🔹 SIFT:
   Keypoints detectados: Img1=1389, Img2=1129
   Total matches: 381
   Distancia promedio (50 mejores): 123.00
   Distancia promedio (todos): 254.66

🔹 ORB:
   Keypoints detectados: Img1=1000, Img2=1000
   Total matches: 241
   Distancia promedio (50 mejores): 36.72
   Distancia promedio (todos): 52.04

{'harris': {'keypoints': (7251, 7799),
  'matches': 645,
  'avg_dist_50': 23.98,
  'avg_dist_all': 43.46201550387597},
 'sift': {'keypoints': (1389, 1129),
  'matches': 381,
  'avg_dist_50': 122.9985629272461,
  'avg_dist_all': 254.66115708238496},
 'orb': {'keypoints': (1000, 1000),
  'matches': 241,
  'avg_dist_50': 36.72,
  'avg_dist_all': 52.04149377593361}}

# Wall-E: Cambio de entorno completo
comparator.compare_algorithms(
    'data/01_Wall-E_Front_Environment1.jpg',
    'data/13_Wall-E_Front_Outside_Environment3.jpg'
)

============================================================
📊 Resultados para: 01_Wall-E_Front_Environment1.jpg vs 13_Wall-E_Front_Outside_Environment3.jpg
============================================================

🔹 HARRIS:
   Keypoints detectados: Img1=7251, Img2=116800
   Total matches: 1782
   Distancia promedio (50 mejores): 24.30
   Distancia promedio (todos): 47.32

🔹 SIFT:
   Keypoints detectados: Img1=1389, Img2=14589
   Total matches: 748
   Distancia promedio (50 mejores): 99.33
   Distancia promedio (todos): 268.20

🔹 ORB:
   Keypoints detectados: Img1=1000, Img2=1000
   Total matches: 242
   Distancia promedio (50 mejores): 35.96
   Distancia promedio (todos): 55.79

{'harris': {'keypoints': (7251, 116800),
  'matches': 1782,
  'avg_dist_50': 24.3,
  'avg_dist_all': 47.32435465768799},
 'sift': {'keypoints': (1389, 14589),
  'matches': 748,
  'avg_dist_50': 99.3261750793457,
  'avg_dist_all': 268.1964029220336},
 'orb': {'keypoints': (1000, 1000),
  'matches': 242,
  'avg_dist_50': 35.96,
  'avg_dist_all': 55.789256198347104}}

# Puzzle: Diferentes vistas y oclusiones
comparator.compare_algorithms(
    'data/15_PopeyePuzzle_Front.jpg',
    'data/16_PopeyePuzzle_MorePuzzles.jpg'
)

============================================================
📊 Resultados para: 15_PopeyePuzzle_Front.jpg vs 16_PopeyePuzzle_MorePuzzles.jpg
============================================================

🔹 HARRIS:
   Keypoints detectados: Img1=16584, Img2=48348
   Total matches: 2408
   Distancia promedio (50 mejores): 18.68
   Distancia promedio (todos): 47.08

🔹 SIFT:
   Keypoints detectados: Img1=2450, Img2=4805
   Total matches: 1100
   Distancia promedio (50 mejores): 51.09
   Distancia promedio (todos): 212.60

🔹 ORB:
   Keypoints detectados: Img1=1000, Img2=1000
   Total matches: 287
   Distancia promedio (50 mejores): 32.24
   Distancia promedio (todos): 57.02

{'harris': {'keypoints': (16584, 48348),
  'matches': 2408,
  'avg_dist_50': 18.68,
  'avg_dist_all': 47.08056478405316},
 'sift': {'keypoints': (2450, 4805),
  'matches': 1100,
  'avg_dist_50': 51.09270275115967,
  'avg_dist_all': 212.59723248741844},
 'orb': {'keypoints': (1000, 1000),
  'matches': 287,
  'avg_dist_50': 32.24,
  'avg_dist_all': 57.01742160278746}}

# Referencia base
base_image = 'data/01_Wall-E_Front_Environment1.jpg'

# Lista de imágenes para comparar
comparison_images = [
    'data/03_Wall-E_Right_Side_Environment2.jpg',
    'data/04_Wall-E_Left_Side_Environment2.jpg',
    'data/06_Wall-E_Close_Up_Front_Environment2.jpg',
    'data/07_Wall-E_Laydown_Environment2.jpg',
    'data/10_Wall-E_And_His_Friends_Free_Front_Environment1.jpg',
    'data/11_Wall-E_And_His_Friends_Right_Side_Behind_Environment1.jpg',
    'data/14_Wall-E_Front_BehindAmongUs_Enviroment2.jpg'
]

# Ejecutar comparaciones
for img_path in comparison_images:
    comparator.compare_algorithms(base_image, img_path)

============================================================
📊 Resultados para: 01_Wall-E_Front_Environment1.jpg vs 03_Wall-E_Right_Side_Environment2.jpg
============================================================

🔹 HARRIS:
   Keypoints detectados: Img1=7251, Img2=40962
   Total matches: 1342
   Distancia promedio (50 mejores): 26.14
   Distancia promedio (todos): 46.79

🔹 SIFT:
   Keypoints detectados: Img1=1389, Img2=6820
   Total matches: 656
   Distancia promedio (50 mejores): 145.63
   Distancia promedio (todos): 274.80

🔹 ORB:
   Keypoints detectados: Img1=1000, Img2=1000
   Total matches: 233
   Distancia promedio (50 mejores): 45.68
   Distancia promedio (todos): 57.71

============================================================
📊 Resultados para: 01_Wall-E_Front_Environment1.jpg vs 04_Wall-E_Left_Side_Environment2.jpg
============================================================

🔹 HARRIS:
   Keypoints detectados: Img1=7251, Img2=37241
   Total matches: 1326
   Distancia promedio (50 mejores): 21.48
   Distancia promedio (todos): 45.71

🔹 SIFT:
   Keypoints detectados: Img1=1389, Img2=6600
   Total matches: 669
   Distancia promedio (50 mejores): 135.13
   Distancia promedio (todos): 273.40

🔹 ORB:
   Keypoints detectados: Img1=1000, Img2=1000
   Total matches: 238
   Distancia promedio (50 mejores): 46.50
   Distancia promedio (todos): 59.34

============================================================
📊 Resultados para: 01_Wall-E_Front_Environment1.jpg vs 06_Wall-E_Close_Up_Front_Environment2.jpg
============================================================

🔹 HARRIS:
   Keypoints detectados: Img1=7251, Img2=30187
   Total matches: 1182
   Distancia promedio (50 mejores): 21.56
   Distancia promedio (todos): 45.18

🔹 SIFT:
   Keypoints detectados: Img1=1389, Img2=5351
   Total matches: 683
   Distancia promedio (50 mejores): 119.29
   Distancia promedio (todos): 264.28

🔹 ORB:
   Keypoints detectados: Img1=1000, Img2=1000
   Total matches: 263
   Distancia promedio (50 mejores): 35.78
   Distancia promedio (todos): 53.68

============================================================
📊 Resultados para: 01_Wall-E_Front_Environment1.jpg vs 07_Wall-E_Laydown_Environment2.jpg
============================================================

🔹 HARRIS:
   Keypoints detectados: Img1=7251, Img2=27800
   Total matches: 1179
   Distancia promedio (50 mejores): 19.58
   Distancia promedio (todos): 45.48

🔹 SIFT:
   Keypoints detectados: Img1=1389, Img2=9459
   Total matches: 750
   Distancia promedio (50 mejores): 113.21
   Distancia promedio (todos): 268.15

🔹 ORB:
   Keypoints detectados: Img1=1000, Img2=1000
   Total matches: 259
   Distancia promedio (50 mejores): 42.06
   Distancia promedio (todos): 57.88

============================================================
📊 Resultados para: 01_Wall-E_Front_Environment1.jpg vs 10_Wall-E_And_His_Friends_Free_Front_Environment1.jpg
============================================================

🔹 HARRIS:
   Keypoints detectados: Img1=7251, Img2=5745
   Total matches: 508
   Distancia promedio (50 mejores): 21.48
   Distancia promedio (todos): 43.71

🔹 SIFT:
   Keypoints detectados: Img1=1389, Img2=937
   Total matches: 411
   Distancia promedio (50 mejores): 90.24
   Distancia promedio (todos): 219.69

🔹 ORB:
   Keypoints detectados: Img1=1000, Img2=1000
   Total matches: 279
   Distancia promedio (50 mejores): 29.14
   Distancia promedio (todos): 47.01

============================================================
📊 Resultados para: 01_Wall-E_Front_Environment1.jpg vs 11_Wall-E_And_His_Friends_Right_Side_Behind_Environment1.jpg
============================================================

🔹 HARRIS:
   Keypoints detectados: Img1=7251, Img2=3706
   Total matches: 381
   Distancia promedio (50 mejores): 26.10
   Distancia promedio (todos): 48.46

🔹 SIFT:
   Keypoints detectados: Img1=1389, Img2=961
   Total matches: 298
   Distancia promedio (50 mejores): 154.76
   Distancia promedio (todos): 267.39

🔹 ORB:
   Keypoints detectados: Img1=1000, Img2=1000
   Total matches: 236
   Distancia promedio (50 mejores): 43.44
   Distancia promedio (todos): 55.72

============================================================
📊 Resultados para: 01_Wall-E_Front_Environment1.jpg vs 14_Wall-E_Front_BehindAmongUs_Enviroment2.jpg
============================================================

🔹 HARRIS:
   Keypoints detectados: Img1=7251, Img2=39092
   Total matches: 1272
   Distancia promedio (50 mejores): 25.82
   Distancia promedio (todos): 49.05

🔹 SIFT:
   Keypoints detectados: Img1=1389, Img2=6663
   Total matches: 637
   Distancia promedio (50 mejores): 163.92
   Distancia promedio (todos): 281.75

🔹 ORB:
   Keypoints detectados: Img1=1000, Img2=1000
   Total matches: 237
   Distancia promedio (50 mejores): 49.76
   Distancia promedio (todos): 61.42

def plot_algorithm_comparison(metrics: dict) -> None:
    """
    Genera gráficos de líneas comparando el rendimiento de los algoritmos.
    """
    fig, (ax1, ax2) = plt.subplots(2, 1, figsize=(16, 12))

    x_labels = metrics['image_pairs']
    x_pos = range(len(x_labels))

    # Gráfico 1: Top 50 matches
    ax1.plot(x_pos, metrics['sift_50'], marker='o', linewidth=2,
             label='SIFT (50 mejores)', color='#3498db')
    ax1.plot(x_pos, metrics['orb_50'], marker='s', linewidth=2,
             label='ORB (50 mejores)', color='#2ecc71')
    ax1.plot(x_pos, metrics['harris_50'], marker='^', linewidth=2,
             label='Harris (50 mejores)', color='#e74c3c')

    ax1.set_xticks(x_pos)
    ax1.set_xticklabels(x_labels, rotation=45, ha='right', fontsize=8)
    ax1.set_ylabel('Distancia Promedio', fontsize=11, fontweight='bold')
    ax1.set_title('Comparación: Distancia Promedio de los 50 Mejores Matches',
                  fontsize=13, fontweight='bold')
    ax1.legend(loc='best', fontsize=10)
    ax1.grid(True, alpha=0.3)

    # Gráfico 2: Todos los matches
    ax2.plot(x_pos, metrics['sift_all'], marker='o', linewidth=2,
             label='SIFT (todos)', color='#1a5490')
    ax2.plot(x_pos, metrics['orb_all'], marker='s', linewidth=2,
             label='ORB (todos)', color='#1d8348')
    ax2.plot(x_pos, metrics['harris_all'], marker='^', linewidth=2,
             label='Harris (todos)', color='#943126')

    ax2.set_xticks(x_pos)
    ax2.set_xticklabels(x_labels, rotation=45, ha='right', fontsize=8)
    ax2.set_ylabel('Distancia Promedio', fontsize=11, fontweight='bold')
    ax2.set_title('Comparación: Distancia Promedio de Todos los Matches',
                  fontsize=13, fontweight='bold')
    ax2.legend(loc='best', fontsize=10)
    ax2.grid(True, alpha=0.3)

    plt.tight_layout()
    plt.show()

# Generar gráfico comparativo
plot_algorithm_comparison(comparator.metrics)

def plot_summary_statistics(metrics: dict) -> None:
    """
    Genera gráfico de barras con promedios generales.
    """
    labels = ['SIFT\n(50)', 'SIFT\n(Todos)', 'ORB\n(50)',
              'ORB\n(Todos)', 'Harris\n(50)', 'Harris\n(Todos)']

    averages = [
        np.mean(metrics['sift_50']),
        np.mean(metrics['sift_all']),
        np.mean(metrics['orb_50']),
        np.mean(metrics['orb_all']),
        np.mean(metrics['harris_50']),
        np.mean(metrics['harris_all'])
    ]

    colors = ['#3498db', '#1a5490', '#2ecc71', '#1d8348', '#e74c3c', '#943126']

    plt.figure(figsize=(12, 7))
    bars = plt.bar(labels, averages, color=colors, edgecolor='black', linewidth=1.2)

    plt.ylabel('Distancia Promedio', fontsize=12, fontweight='bold')
    plt.title('Resumen: Distancia Promedio por Algoritmo (Todas las Comparaciones)',
              fontsize=14, fontweight='bold', pad=20)
    plt.ylim(0, max(averages) * 1.2)
    plt.grid(axis='y', alpha=0.3)

    # Añadir valores sobre las barras
    for bar, value in zip(bars, averages):
        height = bar.get_height()
        plt.text(bar.get_x() + bar.get_width()/2., height,
                f'{value:.2f}',
                ha='center', va='bottom', fontsize=11, fontweight='bold')

    plt.tight_layout()
    plt.show()

plot_summary_statistics(comparator.metrics)

import pandas as pd

def create_comparison_table(metrics: dict) -> pd.DataFrame:
    """
    Crea tabla comparativa con todas las métricas.
    """
    df = pd.DataFrame({
        'Par de Imágenes': [pair.replace('\n', ' ') for pair in metrics['image_pairs']],
        'KP SIFT': metrics['keypoints_sift'],
        'KP ORB': metrics['keypoints_orb'],
        'KP Harris': metrics['keypoints_harris'],
        'Matches SIFT': metrics['matches_sift'],
        'Matches ORB': metrics['matches_orb'],
        'Matches Harris': metrics['matches_harris'],
        'SIFT Dist (50)': [f"{x:.2f}" for x in metrics['sift_50']],
        'ORB Dist (50)': [f"{x:.2f}" for x in metrics['orb_50']],
        'Harris Dist (50)': [f"{x:.2f}" for x in metrics['harris_50']]
    })

    return df

# Crear y mostrar tabla
results_table = create_comparison_table(comparator.metrics)
print("\n" + "="*120)
print("📊 TABLA COMPARATIVA DE RESULTADOS")
print("="*120)
print(results_table.to_string(index=False))
print("="*120)

========================================================================================================================
📊 TABLA COMPARATIVA DE RESULTADOS
========================================================================================================================
                   Par de Imágenes  KP SIFT  KP ORB  KP Harris  Matches SIFT  Matches ORB  Matches Harris SIFT Dist (50) ORB Dist (50) Harris Dist (50)
                  elon_1 vs elon_2      296     967       1125           102          311             152         204.85         33.10            17.26
01_Wall-E_Front vs 02_Wall-E_Right     1389    1000       7251           390          256             618         133.08         39.00            22.80
01_Wall-E_Front vs 08_Wall-E_Upsid     1389    1000       7251           622          246            1135         108.51         42.46            24.44
01_Wall-E_Front vs 05_Wall-E_Zoom_     1389    1000       7251           742          239            1647         169.28         52.40            25.32
01_Wall-E_Front vs 12_Wall-E_Laydo     1389    1000       7251           526          233            1124          95.13         43.80            22.68
01_Wall-E_Front vs 09_Wall-E_And_H     1389    1000       7251           381          241             645         123.00         36.72            23.98
01_Wall-E_Front vs 13_Wall-E_Front     1389    1000       7251           748          242            1782          99.33         35.96            24.30
15_PopeyePuzzle vs 16_PopeyePuzzle     2450    1000      16584          1100          287            2408          51.09         32.24            18.68
01_Wall-E_Front vs 03_Wall-E_Right     1389    1000       7251           656          233            1342         145.63         45.68            26.14
01_Wall-E_Front vs 04_Wall-E_Left_     1389    1000       7251           669          238            1326         135.13         46.50            21.48
01_Wall-E_Front vs 06_Wall-E_Close     1389    1000       7251           683          263            1182         119.29         35.78            21.56
01_Wall-E_Front vs 07_Wall-E_Laydo     1389    1000       7251           750          259            1179         113.21         42.06            19.58
01_Wall-E_Front vs 10_Wall-E_And_H     1389    1000       7251           411          279             508          90.24         29.14            21.48
01_Wall-E_Front vs 11_Wall-E_And_H     1389    1000       7251           298          236             381         154.76         43.44            26.10
01_Wall-E_Front vs 14_Wall-E_Front     1389    1000       7251           637          237            1272         163.92         49.76            25.82
========================================================================================================================

def print_summary_statistics(metrics: dict) -> None:
    """
    Imprime estadísticas resumidas de todos los experimentos.
    """
    print("\n" + "="*80)
    print("📈 ESTADÍSTICAS RESUMIDAS")
    print("="*80)

    algorithms = [
        ('SIFT', 'sift_50', 'sift_all'),
        ('ORB', 'orb_50', 'orb_all'),
        ('Harris', 'harris_50', 'harris_all')
    ]

    for name, key_50, key_all in algorithms:
        print(f"\n🔹 {name}:")
        print(f"   Distancia promedio (50 mejores matches):")
        print(f"      Media: {np.mean(metrics[key_50]):.2f} | "
              f"Desv. Std: {np.std(metrics[key_50]):.2f} | "
              f"Min: {np.min(metrics[key_50]):.2f} | "
              f"Max: {np.max(metrics[key_50]):.2f}")

        print(f"   Distancia promedio (todos los matches):")
        print(f"      Media: {np.mean(metrics[key_all]):.2f} | "
              f"Desv. Std: {np.std(metrics[key_all]):.2f} | "
              f"Min: {np.min(metrics[key_all]):.2f} | "
              f"Max: {np.max(metrics[key_all]):.2f}")

    print("\n" + "="*80)

print_summary_statistics(comparator.metrics)

================================================================================
📈 ESTADÍSTICAS RESUMIDAS
================================================================================

🔹 SIFT:
   Distancia promedio (50 mejores matches):
      Media: 127.10 | Desv. Std: 36.45 | Min: 51.09 | Max: 204.85
   Distancia promedio (todos los matches):
      Media: 261.26 | Desv. Std: 19.71 | Min: 212.60 | Max: 283.92

🔹 ORB:
   Distancia promedio (50 mejores matches):
      Media: 40.54 | Desv. Std: 6.47 | Min: 29.14 | Max: 52.40
   Distancia promedio (todos los matches):
      Media: 55.93 | Desv. Std: 4.32 | Min: 47.01 | Max: 64.68

🔹 Harris:
   Distancia promedio (50 mejores matches):
      Media: 22.77 | Desv. Std: 2.67 | Min: 17.26 | Max: 26.14
   Distancia promedio (todos los matches):
      Media: 45.40 | Desv. Std: 4.80 | Min: 28.60 | Max: 49.22

================================================================================

Propiedad	Descripción
Repetibilidad	El mismo punto debe detectarse en múltiples vistas
Invarianza a Escala	Robusto ante cambios de tamaño del objeto
Invarianza a Rotación	Robusto ante rotaciones de la imagen
Invarianza a Iluminación	Robusto ante cambios de brillo/contraste
Prominencia	Detectar regiones distintivas y salientes
Localización	Features bien localizados espacialmente

Algoritmo	Invarianza Escala	Invarianza Rotación	Velocidad	Descriptor
Harris	❌	✅	⚡⚡⚡	Requiere externo
SIFT	✅	✅	🐢	128D float
ORB	Limitada	✅	⚡⚡⚡	256 bits

🎓 Maestría en Inteligencia Artificial Aplicada¶

🖼️ Visión Computacional para Imágenes y Video¶

👨‍🏫 Profesores¶

Actividad 7.2: Algoritmos de Extracción de Características - Feature Matching¶

📌 Detalles de la Actividad¶

👥 Team 13¶

🚀 Nuestro Equipo¶

Javier Augusto Rebull Saucedo¶

Juan Carlos Pérez Nava¶

Luis Gerardo Sánchez Salazar¶

Oscar Enrique García García¶

Objetivo del Proyecto¶

📚 Tabla de Contenidos¶

🔬 Metodología de Análisis¶

💻 Tecnologías Utilizadas¶

📖 Fundamentos Teóricos: Feature Detection & Matching¶

🎯 El Problema Fundamental¶

🔄 Pipeline de Feature Matching¶

✨ Propiedades Deseadas de Features¶

🔍 Algoritmos Implementados¶

1. Harris Corner Detector (Harris & Stephens, 1988) [6]¶

2. SIFT - Scale Invariant Feature Transform (Lowe, 2004) [7]¶

3. ORB - Oriented FAST and Rotated BRIEF (Rublee et al., 2011) [9]¶

🔗 Feature Matching¶

📊 Comparación Rápida¶

📚 Referencias¶

📑 Tabla de Contenidos¶

1. Importación de Librerías ¶

2. Descarga y Preparación de Imágenes ¶

3. ORB (Oriented FAST and Rotated BRIEF) ¶

4. Creación de Imagen de Prueba con Invarianza ¶

5. Visualización de Imágenes de Entrenamiento y Prueba ¶

6. Detección de Keypoints con ORB ¶

Matching con ORB¶

7. SIFT Matching (Scale Invariant Feature Transform) ¶

8. Matching entre Diferentes Imágenes ¶

9. Extracción de Keypoints con SIFT ¶

10. Feature Matching Comparativo ¶

11. Análisis Comparativo de Algoritmos ¶

Harris Corner Detection¶

Comparación Demo: Elon Musk¶

Comparaciones: Wall-E en Diferentes Condiciones¶

Comparación: Puzzle de Popeye¶

Comparaciones Adicionales: Wall-E¶

Visualización de Resultados Agregados¶

Tabla Comparativa Detallada¶

Estadísticas Agregadas¶

Hallazgos y Análisis¶

Observaciones Principales:¶

Conclusiones sobre Matching:¶

12. Conclusiones Personales ¶

🎓 Javier Augusto Rebull Saucedo (A01795838)¶

🎓 Juan Carlos Pérez Nava (A01795941)¶

🎓 Luis Gerardo Sánchez Salazar (A01232963)¶

🎓 Oscar Enrique García García (A01016093)¶

13. Reflexión Final del Equipo ¶

🚀 Team 13 - Reflexión Colectiva¶

Logros Destacados:¶

Aprendizajes Clave:¶

Aplicaciones Potenciales:¶

Reflexión sobre Trabajo en Equipo:¶

Perspectivas Futuras:¶

📚 Referencias¶